Normal map issue.
polycounter lvl 2
Hi, i have question about my normal map.
After baking the normal map in any program (SP, MARMOSET) there are no artifacts.
But after transferring the normal map to another editor, artifacts appear on normal.
I didn't hard edge the UV seams because it gives bad results for me.
1 screenshot - marmoser(after baking)
2 screenshot - substance painter(green channel inverted)
Replies
Did you triangulate the mesh before baking? If not, my guess is it's due to applications triangulating differently.
Best give just more information (wireframe, UVs, Normalmap) to reduce the guesswork. Ideally attach a problematic part so people can reproduce it on their end.
Hi, thanks for the reply. I triangulated the model and normal map is good now. But I forgot to mention one more problem. When I turn on the metallic, UV seams appear on the model. In any case, I do not need metal on this model, anyway it still worries me.
Padding 32px
Normal map - 4k 16 bit
Well ... first off, why do you have so many hard edges on a lowpoly model that is supposed to be smooth all around ?
Hey, cool that you found a solution!
Looking at your screenshots, I wouldn't necessarily use hard edges on some those UV seams, which are well supported and have a soft curvature. However, I don't know for certain if this would have an effect on the visibility of the seams. Maybe someone else does.
To comment the UVs, one thing you could do is to straighten some UV borders to further reduce potential of visible seams.
After all, I think slight seams visible on curved surfaces under thorough inspection are almost to be expected. It's a smooth surface getting written into a pixel table of a certain resolution into cells with a certain value range. But in practice, those seams often get swallowed by the additional surface details.
Keep it up!
Hi, I was advised to do a hard edge on UV. But this is not important because with a smoothed model the same result is obtained.
Hello,
Well, whoever advised you to do that likely had no idea what they are talking about, as there is zero reason to add an infinitely hard/abrupt surface change at a place where the model is supposed to be smooth. For some reason this myth is still going around after all these years - I suppose because of some people with a poor grasp of english misunderstanding that if an edge is hard, then it should be split in the UVs (which makes total sense) ... and doing the reverse.
Now indeed if this is fully counter-balanced by your nmap then one could say it doesn't matter, but still, better work with a healthy base.
Also ... this thread will soon start running in circles if you don't upload your model.
UV seams don't need a hard edge, but hard edges do need a UV seam.
In my experience the way SP calculates normal details usually gives me a few problems, but since I'm usually exporting to another program for rendering I just ignore it and check the quality regularly in whatever software I'm going to render it. I don't know if this is what you're going to do, but I found that a lot of the problems with normals in SP can just be ignored as long as it looks fine wherever you're exporting it.